The reference carries 44 gases. Anything it does not carry is invisible to the tool — no warning, no finding, just silence.
The clearest example is already documented in gwp_reference.json under not_covered:
HCFC-22 — No row here, so the tool is blind to it. Published: AR4 1,810 / AR5 1,760 / AR6 1,960.
HCFC-22 is still very widely used in older equipment and appears in a lot of refrigerant tables, so this is a real gap rather than a tidy-up.
What to do
Add the gas to values in gwp_reference.json:
"HCFC_22": { "ar4": 1810, "ar5": 1760, "ar6": 1960 }No code change. --self-test verifies the reference's shape, and the gas-name matcher derives its patterns from the keys.
What the PR must say
Where each number came from. A primary source, named: IPCC AR6 Table 7.SM.7, or GHG Protocol "Global Warming Potential Values" v2.0 (August 2024), with the table. "I found it online" is not enough — the whole tool rests on these being right, and a wrong reference value turns into a confident false accusation against someone else's repo.
If two reputable sources disagree for your gas, do not pick one. That is what the exclusions block is for: add the gas, exclude the disputed report, and write the reason. SF₆ AR6 is the worked example — GHG Protocol says 24,300, IPCC says 25,200, and the tool refuses to judge it.
Also check coincides(): if two reports published the same number for your gas, the mislabel test correctly cannot use that pair, and nothing needs doing — but say so in the PR so a reviewer knows you checked.
The reference carries 44 gases. Anything it does not carry is invisible to the tool — no warning, no finding, just silence.
The clearest example is already documented in
gwp_reference.jsonundernot_covered:HCFC-22 is still very widely used in older equipment and appears in a lot of refrigerant tables, so this is a real gap rather than a tidy-up.
What to do
Add the gas to
valuesingwp_reference.json:No code change.
--self-testverifies the reference's shape, and the gas-name matcher derives its patterns from the keys.What the PR must say
Where each number came from. A primary source, named: IPCC AR6 Table 7.SM.7, or GHG Protocol "Global Warming Potential Values" v2.0 (August 2024), with the table. "I found it online" is not enough — the whole tool rests on these being right, and a wrong reference value turns into a confident false accusation against someone else's repo.
If two reputable sources disagree for your gas, do not pick one. That is what the
exclusionsblock is for: add the gas, exclude the disputed report, and write the reason. SF₆ AR6 is the worked example — GHG Protocol says 24,300, IPCC says 25,200, and the tool refuses to judge it.Also check
coincides(): if two reports published the same number for your gas, the mislabel test correctly cannot use that pair, and nothing needs doing — but say so in the PR so a reviewer knows you checked.